AI agents: The new attack surface. A global survey of security, IT professionals and executives

72% of respondents state AI agents pose a greater risk than machine identities.

60% of respondents say AI agent's ability to access privileged data is a factor contributing to AI agents as a security risk.

96% of tech professionals view AI agents as a growing security risk.

98% of organizations plan to expand their adoption of AI agents.

82% of organizations already use AI agents.

Only 44% of organizations report having policies in place to secure AI agents.

57% of respondents say AI agents sharing privileged data is a factor contributing to AI agents as a security risk.

58% of respondents say AI agent's potential to perform unintended actions is a factor contributing to AI agents as a security risk.

Alarmingly, 23% reported their AI agents have been tricked into revealing access credentials.

39% of respondents say AI agents accessed unauthorized systems or resources.

31% of respondents say AI agents accessed inappropriate data.

54% of respondents say AI agents accessing and sharing inappropriate information is a factor contributing to AI agents as a security risk.

55% of respondents say AI agents making decisions based on inaccurate or unverified data is a factor contributing to AI agents as a security risk.

80% of companies say their AI agents have taken unintended actions.

An overwhelming 92% state that governing AI agents is critical to enterprise security.

32% of respondents say AI agents downloaded sensitive content.

60% of respondents say AI agent's ability to access privileged data is a factor contributing to AI agents as a security risk.

58% of respondents say AI agent's potential to perform unintended actions is a factor contributing to AI agents as a security risk.

57% of respondents say AI agents sharing privileged data is a factor contributing to AI agents as a security risk.

54% of respondents say AI agents accessing and sharing inappropriate information is a factor contributing to AI agents as a security risk.

39% of respondents say AI agents accessed unauthorized systems or resources.

31% of respondents say AI agents accessed inappropriate data.

55% of respondents say AI agents making decisions based on inaccurate or unverified data is a factor contributing to AI agents as a security risk.